Megabase-scale copy number variants (CNVs) can have profound phenotypic consequences. Germline CNVs of this magnitude are associated with disease and experience negative selection. However, it is unknown whether organismal function requires that every cell maintain a balanced genome. It is possible that large somatic CNVs are tolerated or even positively selected. Single-cell sequencing is a useful tool for assessing somatic genomic heterogeneity, but its performance in CNV detection has not been rigorously tested. Here, we develop an approach that allows for reliable detection of megabase-scale CNVs in single somatic cells. We discover large CNVs in 8%-9% of cells across tissues and identify two recurrent CNVs. We conclude that large CNVs can be tolerated in subpopulations of cells, and particular CNVs are relatively prevalent within and across individuals.[Supplemental material is available for this article.]Copy number variants (CNVs) can range in size from hundreds to millions of base pairs. Copy number changes affect approximately seven times as many base pairs as single-nucleotide variants and are major contributors to inter-individual differences (Sudmant et al. 2015). More than 65% of individuals harbor a germline CNV of at least 100 kb, and at least 1% of individuals have a CNV exceeding 1 Mb (Itsara et al. 2009). Although megabase-scale CNVs could be considered collectively common, the specific CNVs themselves are rare and often associated with disease (Girirajan et al. 2011). Not surprisingly, large CNVs experience negative selection, and their existence in a population is largely due to de novo events (Itsara et al. 2010).Although germline, megabase-scale CNVs are found in 1% of individuals, the prevalence of somatic CNVs is only beginning to be investigated. Array-based analyses of populations of cells from many individuals provided initial insight into this question. These studies identified megabase-scale somatic aberrations in up to 4% of individuals; however, the sensitivity was limited to CNVs present in >5% of cells (Forsberg et al. 2012;Jacobs et al. 2012;Laurie et al. 2012). These studies are thus blind to alterations that arise late in development or adversely affect fitness, as this would limit their propagation in a cell population. With the emergence of methods to amplify the genome of a single cell, single-cell sequencing now provides an alternate means of assessing the prevalence of somatic CNVs and offers the advantage of detecting variants that exist in as few as one cell. Recently, two groups performed low-coverage sequencing of single human neurons and reported at least one megabase-scale CNV in >40% of neurons (McConnell et al. 2013;Cai et al. 2014). These findings suggest much greater tolerance of large somatic CNVs compared to germline CNVs and raise the interesting possibility that somatic genomic heterogeneity contributes to phenotypic diversity within a tissue. However, it is still unclear how CNV detection methods perform when applied to individual cells, as single-cell sequ...